Feedcore: Improve parsing of publishing date

This works around problems with invalid date formats
and also prefers the "published" field over "updated".

See also Feedparser issue 327:
http://code.google.com/p/feedparser/issues/detail?id=327

src/gpodder/feedcore.py

@@ -24,6 +24,13 @@
 
 import feedparser
 
+try:
+    # Python 2
+    from rfc822 import mktime_tz
+except ImportError:
+    # Python 3
+    from email.utils import mktime_tz
+
 def patch_feedparser():
     """Monkey-patch the Universal Feed Parser"""
     # Detect the 'plain' content type as 'text/plain'
@@ -262,3 +269,35 @@ def fetch(self, url, etag=None, modified=None):
         """
         self._parse_feed(url, etag, modified)
 
+
+def get_pubdate(entry):
+    """Try to determine the real pubDate of a feedparser entry
+
+    This basically takes the updated_parsed value, but also uses some more
+    advanced techniques to work around various issues with ugly feeds.
+
+    "published" now also takes precedence over "updated" (with updated used as
+    a fallback if published is not set/available). RSS' "pubDate" element is
+    "updated", and will only be used if published_parsed is not available.
+    """
+
+    pubdate = entry.get('published_parsed', None)
+
+    if pubdate is None:
+        pubdate = entry.get('updated_parsed', None)
+
+    if pubdate is None:
+        # See http://code.google.com/p/feedparser/issues/detail?id=327
+        updated = entry.get('published', entry.get('updated', None))
+        if updated is not None:
+            # FIXME: This is kludgy. We should write our own date handler
+            # and register it with feedparser.registerDateHandler() and/or
+            # wait for feedparser to add support for this bogus date format.
+            pubdate = feedparser._parse_date(updated.replace(',', ''))
+
+    if pubdate is None:
+        # Cannot determine pubdate - party like it's 1970!
+        return 0
+
+    return mktime_tz(pubdate + (0,))
+

src/gpodder/model.py

@@ -41,13 +41,6 @@
 import time
 import datetime
 
-try:
-    # Python 2
-    from rfc822 import mktime_tz
-except ImportError:
-    # Python 3
-    from email.utils import mktime_tz
-
 import hashlib
 import feedparser
 import collections
@@ -218,8 +211,7 @@ def from_feedparser_entry(cls, entry, channel):
         if not episode.description:
             episode.description = entry.get('subtitle', '')
 
-        if entry.get('updated_parsed', None):
-            episode.published = mktime_tz(entry.updated_parsed+(0,))
+        episode.published = feedcore.get_pubdate(entry)
 
         enclosures = entry.get('enclosures', [])
         media_rss_content = entry.get('media_content', [])
@@ -1011,8 +1003,7 @@ def _consume_updated_feed(self, feed, max_episodes=0):
             # max_episodes old episodes, new episodes will not be shown.
             # See also: gPodder Bug 1186
             try:
-                entries = sorted(feed.entries, \
-                        key=lambda x: x.get('updated_parsed', (0,)*9), \
+                entries = sorted(feed.entries, key=feedparser.get_pubdate,
                         reverse=True)[:max_episodes]
             except Exception, e:
                 logger.warn('Could not sort episodes: %s', e, exc_info=True)